aHey Product Hunt! 👋 I built Daena-Coder because I was tired of "one model, one guess" coding tools. The idea: what if you ran 3 LLMs from completely different training families in parallel — Qwen (Alibaba), DeepSeek, and Llama (Meta) — and synthesized the best answer from all three? That's COUNCIL mode, the default. For harder problems, QUINTESSENCE mode adds 5 expert lenses (Architect, Security, Practitioner, Critic, Researcher) each analyzing the responses independently — then meta-synthesizes everything into one authoritative answer. Everything runs locally via Ollama. Your code never leaves your machine. No subscriptions, no API keys required. It also ships with: - 10 built-in slash-command skills (/review, /security, /debug, /refactor...) - Native MCP support — connect GitHub, Postgres, Slack, any MCP server - TURBO mode for 8GB VRAM machines (3 small models truly in parallel) GitHub: https://github.com/Mas-AI-Offici... Happy to answer any questions! This is the open-source coding layer — the full governed platform lives at daena.mas-ai.co

Daena-Coder

Multi-LLM coding desktop - council synthesis, runs locally

Multi-LLM coding desktop - council synthesis, runs locally